The current traffic system allocates a static time to green-light/red-light traffic signals at cross-roads which may lead to problems like waiting for a longer time for your turn even though there are no/fewer vehicles at other lanes. Our proposed system detects vehicle density/no. of vehicles on each lane and based on that, predicts time and dynamically allocates green-light window time to that lane, thus tackling the previously mentioned problem. Also, it detects any unwanted congestions like accident or any traffic jams in any lane and generates a message for the same.
The biggest challenge was detecting the vehicles present on the lanes during different conditions, and also keeping all the lanes in sync, i.e. calculating the accurate time for each without disturbing the other lanes.
For the detection problem, we used trained datasets which gave almost 80-90% accurate results.
Technologies used
Discussion